Working Model [WM2D] - just how does it compute kinematics ?

[fletcher]

A request for help ? ..

Any contribution or assistance from knowledgeable forum members will be appreciated, especially programmers who have "looked behind the scenes" at WM2d's functionality, matrices & code etc.

I am a long time WM user preferring it to real world builds to quickly test object interactions etc - I find it invaluable for the task.

I have no programming experience but have always felt that the program could under the right circumstances show an anomalous kinetic energy gain with the right arrangement of parts & interactions - having said that I am well aware of the conservation laws, thermodynamics & Noether's Theorem etc, but still I persist in exploring unique arrangements that might fit the bill.

To that end I have always assumed that the WM program worked on a bottom up basis rather than a top to bottom trickle down process - by that I mean that each constituent part of physics formulas feed into the resultant kinematics on screen as a building block for the next time frame interval etc - I don't/didn't believe that there was an overarching requirement in the programing to override & keep everything inside the parameters of conservation laws etc e.g. CoE & CoM & CoAM - perhaps I am mistaken in that belief ? - some certainly think so.

However, I know that the program will allow me to get an OU result if I use a fictitious force application at the appropriate point of application - so it seems that I do not get a drop box that says that this operation violates the "ethics' of the program & can not be done ?!

Please read the contributions from two forum members below at BW.com on the subject - I would like a definitive answer to this question - please don't bomb me with advice to not pursue something that violates Conservations Laws - I know the arguments quite well - thanks in advance for a reasoned argument.

http://www.besslerwheel.com/forum/viewtopic.php?p=124796#124796

http://www.besslerwheel.com/forum/viewtopic.php?p=124804#124804

P.S. broli once had some interesting & technical observations about the programming & its functionality - I respect his opinion in most things he says & that was no different - it would be interesting to revisit it.

[helloha]

how does it compute kinematics ?

i'm also interested to know how, any help .... tutorials ?

[LibreEnergia]

A request for help ? ..

Any contribution or assistance from knowledgeable forum members will be appreciated, especially programmers who have "looked behind the scenes" at WM2d's functionality, matrices & code etc.

I am a long time WM user preferring it to real world builds to quickly test object interactions etc - I find it invaluable for the task.

I have no programming experience but have always felt that the program could under the right circumstances show an anomalous kinetic energy gain with the right arrangement of parts & interactions - having said that I am well aware of the conservation laws, thermodynamics & Noether's Theorem etc, but still I persist in exploring unique arrangements that might fit the bill.

To that end I have always assumed that the WM program worked on a bottom up basis rather than a top to bottom trickle down process - by that I mean that each constituent part of physics formulas feed into the resultant kinematics on screen as a building block for the next time frame interval etc - I don't/didn't believe that there was an overarching requirement in the programing to override & keep everything inside the parameters of conservation laws etc e.g. CoE & CoM & CoAM - perhaps I am mistaken in that belief ? - some certainly think so.

However, I know that the program will allow me to get an OU result if I use a fictitious force application at the appropriate point of application - so it seems that I do not get a drop box that says that this operation violates the "ethics' of the program & can not be done ?!

Please read the contributions from two forum members below at BW.com on the subject - I would like a definitive answer to this question - please don't bomb me with advice to not pursue something that violates Conservations Laws - I know the arguments quite well - thanks in advance for a reasoned argument.

http://www.besslerwheel.com/forum/viewtopic.php?p=124796#124796

http://www.besslerwheel.com/forum/viewtopic.php?p=124804#124804

P.S. broli once had some interesting & technical observations about the programming & its functionality - I respect his opinion in most things he says & that was no different - it would be interesting to revisit it.

I found some information about the maths behind it at http://www.maelabs.ucsd.edu/cosmos/resources/wm2d/documentation

I don't know if it still represents the actual algorithms used. However if it does, it would appear that it favours computational speed over accuracy. It uses simplified equations of motion rather than finite element methods. Also it does not truly model flexible bodies but instead considers them to be rigid bodies connected by springs.

Such simulations have their place, but to my mind the only 'true' physics simulation would involve non-linear finite element methods. This allows the simulation of non-rigid multiple body systems with an accuracy only limited by computational effort

No amount of coaxing will make a correct FEM model over-unity. The total sum of all the energies for the system between one state and another is constrained to have zero difference. Any deviation from that is simply an artefact of the numerical methods used to solve the system

[fletcher]

I found some information about the maths behind it at http://www.maelabs.ucsd.edu/cosmos/resources/wm2d/documentation

I don't know if it still represents the actual algorithms used. However if it does, it would appear that it favours computational speed over accuracy. It uses simplified equations of motion rather than finite element methods. Also it does not truly model flexible bodies but instead considers them to be rigid bodies connected by springs.

Such simulations have their place, but to my mind the only 'true' physics simulation would involve non-linear finite element methods. This allows the simulation of non-rigid multiple body systems with an accuracy only limited by computational effort

No amount of coaxing will make a correct FEM model over-unity. The total sum of all the energies for the system between one state and another is constrained to have zero difference. Any deviation from that is simply an artefact of the numerical methods used to solve the system

Thanks for your input Libre - the sim accuracy can be changed from fast-approximate Euler to accurate Kutta-Merson if desired - all sims are a compromise as are many programs - the ultimate sim is a real world build.

There are times when small numerical errors give a false positive as well as known bugs that require alternate mechanical methods of implementation as a work-around - generally these do not concern an experienced user.

Perhaps I have been using this program to long ?! - it seems very logical to me & as the documentation you provided said [thanks, that is a valuable resource] it uses constraint force algorithms, then finds the accelerations, then applies the laws of motion [constraint equations] & applies to solving matrices etc - sometimes I do a check & do the math long hand & find I use the same logical approach as the program [only simplified for point masses for example] - so it seems an intuitive approach to me.

Nowhere did I read that it was constrained to the bounds of conservation laws in a trickle down approach though one could argue it could be implied from first principles that it must be the case ?! - in fact it reads to me that it builds a model basic block by basic block starting with forces & accelerations.

But then, that is why I am asking the question !

ETA:

I re-read the posts on BW.com - I'm not convinced that the kinematics program is constrained to Conservation of Energy &/or Momentum - jim_mich seems to think that they are the 'go to' equations & a solver algorithm finds a 'best fit' scenario where the sum of the energy gain or loss is zero or thereabouts - I know we often take this approach when doing long hand analysis of systems - it is convenient & expedient most times, especially with spreadsheets [e.g. Bernoulli fluid dynamics] - however, if you are looking for a special arrangement or interaction that would defy or negate first principles in certain circumstances then this top down circular logic approach would be erroneous, IMO.

I find WM very reliable - in the early years I built sim facsimiles & real world counterparts to test the sim against - the sim would be tweaked with coefficients of elasticity, moments of inertia, friction coefficients, air drag etc to get a very close performance match indeed - that gave me confidence, & the fact it is used in industry, that its reliability & predictability factors are trustworthy in almost every instance - to add context: in my latest iterations of a device I have cause to question how the program is calculating & functioning because my logic & intuition for once seems to disagree with the program - yet, it is too complex for me to long hand reliably without circular logic - I know I am not infallible but is the programming ?